Search

[Display omitted] • SOC storage potential in cropland topsoil was estimated by a data-driven approach. • Carbon landscape systems were determined by mixture model clustering analysis. • Drivers of SOC storage potential varies in carbon landscape systems. • Cropping and soil managements explained ∼20 % of SOC storage potential variability. • Conversion to rice-based cropping systems can partly achieve 4p1000 initiative. Soil organic carbon (SOC) is receiving increasing attention due to its large storage potential in global carbon cycles and its great importance to soil fertility, agricultural production, and ecosystem services. The increases of SOC storage and reliable estimation of its potential are essential for evaluating the soil sustainability and climate change adaptation under intensive cultivation. In this work, a data-driven approach combining mixture clustering and Random Forest models was proposed to estimate the SOC storage potential of cropland topsoil and its controlling factors in East China. The carbon landscapes systems (CLSs) were delineated using a mixture clustering model by combining the climatic condition, soil properties, cropping systems, and soil management practices. The SOC storage potentials with 95 % confidence intervals at 250 m spatial resolution were estimated as the difference between the current SOC stock and empirically maximum SOC stock at basic (75 %), intermediate (85 %), and ambitious (95 %) expectation objectives for each CLS. The SOC storage potential increased with the increasing of expectation objective settings, with the averaged levels of 13.1, 20.8, and 35.5 t C ha−1 at 75 %, 85 %, and 95 % percentile objectives, respectively. The variable importance from Random Forest indicated that the cropping systems and soil management practices were the unignorable factors controlling the SOC storage potential beyond the climatic conditions and soil properties. Moreover, the shifts of human-induced controlling factors, e.g., cropping systems, also indicated their capability of SOC sequestration potential for partly achieving the "4p1000" initiative (annual growth rate of 0.4 % carbon stocks in the first 30 cm of topsoil). The currently optimal soil management practices for achieving the SOC sequestration potential was the combination of rice-based cropping systems, straw return, and organic fertilizer applied. The data-driven approach coupling with CLSs improved our understanding of the controlling factors on SOC storage potential at regional level with homogenous conditions, enabling evidence-based decision making in promoting carbon sequestration by adopting locally feasible soil management practices. [ABSTRACT FROM AUTHOR]

• In total, 29,927 topsoil samples were collected across Zhejiang province. • Random forest showed good predictive performance with R 2 of 0.65. • Current status of soil TN and C/N ratio were mapped at 250 m resolution. • Mild increase trend of soil TN with slight decline of C/N ratio during 1979–2008. • Intensifed N fertilization contributed to increase of soil TN and decrease of C/N. Adequate soil total nitrogen (TN) is critical to crop productivity, and soil carbon-to-nitrogen ratio (C/N) is an important indicator of soil fertility. Accurate knowledge of soil TN content and C/N ratio is essential for precision agriculture and soil biogeochemical modeling. This study aims to characterize the spatial and temporal trends of soil TN level and C/N ratio for cropland soil across Zhejiang province of East China. A total of 29,927 topsoil (0–20 cm) samples were collected during 2007–2008 for analyzing soil properties, on the basis of which we mapped the soil TN content at a 250 m resolution using a random forest model (RF). With 22 predictors covering soil properties, fertilization, meteorology, locations, and population density, the RF model showed good predictive performance with cross-validation R 2 = 0.65 and RMSE = 0.43 g kg−1. The soil pH, cation exchange capacity (CEC), and fertilization played important roles in determining the spatial variations of soil TN and C/N ratio. The overall level of cropland soil TN within the study area was estimated to be 1.75 ± 0.43 g kg−1 (μ ± σ), varying from 0.25 to 3.64 g kg−1. By comparing our estimates with the previous extensive survey for 1979–1985, we found that the soil TN has slightly increase trend and the soil C/N ratio slightly decreased, which was attributed to intensified N fertilization in the last decade. This work provides up-to-date knowledge of soil TN content and soil C/N ratio in Zhejiang province of East China, indicating the importance of updating soil property databases for better regional cropland management. [ABSTRACT FROM AUTHOR]

Soil organic carbon (SOC) is important to soil fertility and the global carbon cycle. Accurate estimates of SOC stock and its dynamics are critical for managing agricultural ecosystems and carbon accounting under climate change, especially for highly cultivated regions. We extensively surveyed the SOC levels in 29,927 sites in Zhejiang province, an intensively cultivated region of East China, from year 2007 to 2008. We then estimated the spatial distribution of topsoil (0–30 cm) organic carbon stock using a random forest (RF) model, which is a powerful machine learning algorithm with superior predictive performance over parametric statistical models. The final RF model contained 23 predictor variables, covering soil properties, vegetation, climate, topography, land cover, farming practices, and locations. The RF model showed high performance in predicting the SOC stock, with a coefficient of determination ( R 2 ) of 0.76 and a root mean square error (RMSE) of 10.63 t C ha −1 . This performance was superior to the General Linear Model (GLM) ( R 2 = 0.35, RMSE = 19.93 t C ha −1 ) and the ordinary kriging (OK) method ( R 2 = 0.57, RMSE = 14.44 t C ha −1 ), and was equivalent to Boosted Regressing Trees (BRT) ( R 2 = 0.73, RMSE = 11.26 t C ha −1 ). According to the variable importance evaluation, soil properties were the most important predictor variables, followed by climate and location, with relative importance values of 61%, 17%, and 14%, respectively. The predicted SOC stock ranged from 14.8 to 125.5 t C ha −1 , with an average ± standard deviation of 50.1 ± 12.3 t C ha −1 . The mean SOC level obtained from this survey was considerably lower than the value of 60.5 t C ha −1 reported for the same region in the Harmonized World Soil Database (HWSD), which is the most commonly used soil database worldwide. A large spatial discrepancy of SOC stock was observed between this survey and HWSD in regional and sub-regional levels. This study provided an updated regional baseline map of SOC levels for improving farmland management and refining carbon accounting under climate change. [ABSTRACT FROM AUTHOR]

As a rapid, inexpensive and accurate analysis technique, vis–NIR spectra has shown great advantages for determining a wide variety of soil properties, such as soil organic matter content, mineral composition, water content, particle size and color. Thus, this technique is becoming increasingly popular in soil science. We aim to assess the applicability of using vis–NIR spectra to estimate eighteen different soil properties that are important for Chinese Soil Taxonomy (CST). In this study, vis–NIR reflectance spectra were measured under laboratory conditions. First, partial least-squares regression (PLSR) was used to predict the following soil properties related to soil classification: coarse crumb, sand, silt, and clay contents, bulk density (BD), pH (H 2 O), pH (KCl), soil organic matter (SOM), total nitrogen (TN), total potassium (TK), and total phosphorus (TP) contents, cation exchange capacity (CEC), free iron (Fe 2 O 3 ), soluble salts (salt), available phosphorus (AP), exchangeable aluminum (ExAl), aluminum saturation (AS) and base saturation (BS). Then, the important bands for modeling these soil properties were selected based on the selectivity ratio (SR). Finally, the spectral chromophores of the soil and the correlations of soil properties were analyzed. The results showed that (1) the prediction accuracy based on the PLSR algorithm was good for pH, SOM, TN, Fe 2 O 3 , salt, AS and BS (RPD > 2.0, R 2 between 0.70 and 0.90). For sand, silt, clay, BD, TP, TK, CEC, AP and ExAl, the PLSR model could achieve acceptable estimates (1.4 < RPD < 2.0, R 2 between 0.56 and 0.72), while for coarse crumb, the PLSR model was unable to make reliable predictions (RPD < 1.4, R 2 below 0.50). (2) As chromophore properties, SOM, TN, Fe 2 O 3 , clay and salt are and can be predicted by spectroscopy. Besides, BD, pH, TK, TP, CEC, AP, ExAl, AS and BS have significant correlations with chromophore properties and can also be predicted by vis–NIR spectroscopy. Therefore, except for coarse crumb, the soil properties important to CST can be quantitatively predicted by PLSR based on vis–NIR reflectance spectroscopy. This study is significant to CST, and it provides a fast and efficient method for soil classification. [ABSTRACT FROM AUTHOR]

Land reclamation has been highly intensive in China, resulting in a large amount of soil organic carbon (SOC) loss to the atmosphere. Evaluating the factors which drive SOC dynamics and carbon sequestration potential in reclaimed land is critical for improving soil fertility and mitigating global warming. This study aims to determine the current status and factors important to the SOC density in a typical reclaimed land located in Eastern China, where land reclamation has been undergoing for centuries. A total of 4746 topsoil samples were collected from 2007 to 2010. The SOC density of the reclaimed land (3.18 ± 0.05 kg C m − 2 ; mean ± standard error) is significantly lower than that of the adjacent non-reclaimed land (5.71 ± 0.04 kg C m − 2 ) ( p < 0.05). A Random Forest model is developed and it captures the relationships between the SOC density and the environmental/anthropogenic factors ( R 2 = 0.59). The soil pH, land use, and elevation are the most important factors for determining SOC dynamics. In contrast, the effect of the reclamation age on the SOC density is negligible, where SOC content in the land reclaimed during years 1047–1724 is as low as that reclaimed during years 1945–2004. The scenario analysis results indicate that the carbon sequestration potential of the reclaimed lands may achieve a maximum of 5.80 ± 1.81 kg CO 2 m − 2 (mean ± SD) when dryland is converted to flooded land with vegetable-rice cropping system and soil pH of ~ 5.9. Note that in some scenarios the methane emission substantially offsets the carbon sequestration potential, especially for continuous rice cropping system. With the optimal setting for carbon sequestration, it is estimated that the dryland reclaimed in the last 50 years in China is able to sequester 0.12 million tons CO 2 equivalent per year. [ABSTRACT FROM AUTHOR]

A field experiment with five lime rates (0, 3.75, 7.50, 11.25, and 15.00 Mg ha-1) was maintained on a red soil (Ultisol) for 15 years to determine changes of soil acidity and effect on crop yields. The soil acidity decreased while exchangeable Ca in plough layer (0-20 cm) increased with lime rate and time. The decreased subsoil (20-60 cm) acidity started to occur four years after liming, and the extent of decreased soil acidity increased with lime rate and time. The increased ranges of exchangeable Mg2+ in subsoil were greater than that of exchangeable Ca2+, suggesting that downward movement of Mg2+ into the subsoil was faster than that of Ca2+. Lime application significantly increased the yields of crops studied. During the period of experiment, the maximum yield increased 4.67 times for barley, 2.24 times for mungbean, 57.3% for wheat, 53.4% for sesame, 52.8% for broad bean, 44.1% for potato, 35.1% for rapeseed, 32.1% for cotton, 28.4% for corn, 18.5% for watermelon, 11.0% for cowpea and 8.8 % for soybean. Liming at the highest rate was in favor to the decline in subsoil acidity and yield increase, especially for the later period of the experiment. Residual effects of reducing soil acidity by liming for the treatments of 0.5 L, 1.0 L, 1.5 L, and 2.0 L could last for 5, 7, 12, and 14 years, respectively, and the effects of lime application on the yields could last for more than 15 years. [ABSTRACT FROM AUTHOR]

Wise soil management requires detailed soil information, but conventional soil class mapping in a rather coarse spatial resolution cannot meet the demand for precision agriculture. With the advantages of non-destructiveness, rapid cost-efficiency, and labor savings, the spectroscopic technique has proved its high potential for success in soil classification. Previous studies mainly focused on predicting soil classes using a single sensor. In this study, we attempted to compare the predictive ability of visible near infrared (vis-NIR) spectra, mid-infrared (MIR) spectra, and their fused spectra for soil classification. A total of 146 soil profiles were collected from Zhejiang, China, and the soil properties and spectra were measured by their genetic horizons. Along with easy-to-measure auxiliary soil information (soil organic matter, soil texture, color and pH), four spectral data, including vis-NIR, MIR, their simple combination (vis-NIR-MIR), and outer product analysis (OPA) fused spectra, were used for soil classification using a multiple objectives mixed support vector machine model. The independent validation results showed that the classification model using MIR (accuracy of 64.5%) was slightly better than that using vis-NIR (accuracy of 64.2%). The predictive model built on vis-NIR-MIR did not improve the classification accuracy, having the lowest accuracy of 61.1%, which likely resulted from an over-fitting problem. The model based on OPA fused spectra performed best with an accuracy of 68.4%. Our results prove the potential of fusing vis-NIR and MIR using OPA for improving prediction ability for soil classification. [ABSTRACT FROM AUTHOR]

Reliable origin authentication methods are critical for protecting high-value food products with designated geographical origins. A total of 623 tea samples were collected from important green tea production regions around China from 2012 to 2016. A Random Forest model (RF) with 19 input predictors (e.g., δ 13C, 24Mg, 85Rb, and 206Pb/207Pb) was developed. Our RF model not only discriminated Westlake Xihu Longjing green tea (XHLJ) from other regions with an accuracy of 97.6%, but also correctly identified green tea from surrounding regions with an accuracy of 97.9%. The geographical discrimination of tea subsequently harvested in the following years also showed good reliability. Predictive accuracies were higher than 91%. 85Rb, 24Mg, δ 13C and 39K were the most important geographical proxies for determining geographical origin of tea with a relative contribution of 20.6%, 12.5%, 12.1% and 7.4%, respectively. This RF model showed higher classification accuracy than other commonly used chemometrics models and provides a new insight into the use of predictive models utilizing historical data for geographical authentication of agricultural products with Protected Designation of Origin (PDO). • XHLJ tea was differentiated from tea produced in nearby origins by random forest. • Multiple year data strengthened the prediction reliability of data for future years. • Random forest showed superior predictive ability than PCA-based models. • 85Rb, 24Mg, δ13C and 39K were important contributors for tea origin differentiation. [ABSTRACT FROM AUTHOR]